Marketing Dashboard Build and Review

List every system feeding the dashboard: GA4, the CRM (HubSpot or Salesforce), the MAP (Marketo, HubSpot, Pardot), ad platforms (Google Ads, Meta, LinkedIn, TikTok), the CDP if present (Segment, mParticle), and any spreadsheets still in the loop. Record the connector type and refresh method for each — native connector, Supermetrics/Fivetran, or manual export.

Pull the last 30 days of source/medium/campaign values from GA4 and check against the team's UTM convention doc. Casing drift (Email vs email), inconsistent campaign names, and missing utm_content are the usual culprits — they make channel reports uncomparable. Fix at the source by updating the link-builder, not by remapping in the dashboard.

Walk every conversion event in GA4 against the actual user action — common failure mode is a sign-up event firing on email-blur instead of submit, inflating reported conversions by 3-5x. Use GA4 DebugView with Tag Assistant to confirm each event fires once per real conversion.

Connect through the dashboard tool of record — Looker Studio, Tableau, Power BI, or the BI layer over the warehouse. Prefer warehouse-backed connections (BigQuery, Snowflake) over direct API connectors when row counts approach platform caps; native connectors silently sample at high volumes.

Set refresh cadence per source: paid-media platforms hourly during active campaigns, GA4 every 4 hours, CRM nightly. Add a freshness indicator on each tile so viewers can see when data last loaded — stale-data confusion is the most common driver of distrust in the dashboard.

Every KPI on the dashboard must trace to a marketing OKR or pipeline target. Drop vanity metrics (raw impressions, raw followers) unless they're explicitly part of a brand objective. The CMO should be able to point to any tile and name the OKR it supports.

Lock the scoring thresholds with RevOps before publishing. Confirm the lead score that promotes to MQL, the SDR-acceptance criteria for SAL, and the opportunity-creation rules for SQL. Mismatches between marketing's MQL definition and sales's SAL definition are the most-cited source of pipeline disagreement at QBRs.

Pick one model and document the choice on the dashboard's About tile. Last-touch undercredits TOFU content; first-touch undercredits BOFU. Position-based or data-driven splits the difference but requires sufficient conversion volume. Note the model and the lookback window so reviewers can read the numbers correctly.

Each KPI gets a named owner — demand-gen lead owns pipeline, content manager owns organic sessions, lifecycle marketer owns email engagement. The owner is who answers questions about the number at the WBR, not who runs the dashboard.

Open with a top strip of headline KPIs (pipeline, MQLs, ROAS) so a 10-second skim answers the CMO's first question. Group below by channel — paid, organic, email, social — then by funnel stage. Keep one screen for the WBR view; secondary detail goes on linked tabs.

Use a sparkline or 13-week trend on every headline KPI — single-number tiles without history hide regressions. Funnel chart shows session → lead → MQL → SQL → opportunity conversion rates so the WBR conversation can target the leakiest stage.

Confirm every tile renders cleanly at 1440px and on a mobile viewport — Looker Studio mobile layouts diverge from desktop and routinely clip legends. Check accessibility basics: color contrast, no red/green-only encodings, alt text on logo images.

Walk the CMO, demand-gen lead, content lead, and RevOps through the dashboard live. Capture every requested change in the reviewer-notes field below — do not promise verbal-only changes. Approval blocks publish; capture this on the record.

Address every item in the reviewer-notes field, then re-circulate by email with a changelog. Do not republish silently — stakeholders need to know what moved before the next WBR.

Set Slack or email alerts on the metrics that change strategy: pipeline pacing below 80% of target, ROAS dropping below break-even, MQL volume off baseline by more than 25%. Avoid alerting on every tile — alert fatigue kills the channel within a week.

Configure a Monday-morning PDF or PNG export to the marketing-leadership channel ahead of the weekly business review, plus a first-of-month export for the MBR. Put the export on a service account so it doesn't break when an individual leaves.

Walk the dashboard tile by tile. Anything that's a real anomaly — not a known seasonal dip — gets flagged for investigation before the WBR. Common false positives: a campaign launch day, a one-off PR spike, a tracking-pixel deploy that double-counted briefly.

Drill into the source data, not just the dashboard tile. Check GA4 raw events, the CRM record-level export, and the ad-platform UI for the same window. Anomalies that turn out to be tracking artifacts get a fix ticket; anomalies that turn out to be real become a WBR talking point.

Log the takeaways and any spend or strategy shifts in the dashboard's decisions-log tile. The point isn't archiving — it's building a record so the next quarter's planning conversation can reference what worked.

Marketing team gets edit on their own tabs, view on others. Sales leadership gets view on the pipeline tabs only. Avoid sharing edit access broadly — accidental tile deletes are the top support ticket. Use Google Groups or the BI tool's group feature so leavers don't require manual cleanup.

Verify the CMP (OneTrust, Cookiebot, Didomi) blocks the Meta Pixel, LinkedIn Insight Tag, and GA4 from firing pre-consent. The dashboard's traffic numbers should match consented-traffic only — if Meta Pixel is loading before the banner, that's both a privacy violation and a number that won't reconcile with the CMP report.

Confirm GA4 user-data retention is set to the documented retention window (default is 2 months; legal may require 14 months or a different ceiling). Check that DSAR-deleted records have flowed through to the warehouse and dashboard, and that EU and California traffic are processed under the documented lawful basis.

Export the dashboard's user list, cross-reference against current employees and active vendors, and remove anyone who's left or whose engagement has ended. Agency contractors are the most common stale-access source; put their access on a quarterly expiry by default.